spatial data modelling for 3d gis · 1.4 problems associated with spatial modelling for 3d gis 9...

29
Spatial Data Modelling for 3D GIS

Upload: others

Post on 18-Aug-2020

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Spatial Data Modelling for 3D GIS

Page 2: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Alias Abdul-Rahman · Morakot Pilouk

Spatial Data Modellingfor 3D GIS

ABC

Page 3: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Library of Congress Control Number:

ISBN 978-3-540-74166-4 Springer Berlin Heidelberg New York

This work is subject to copyright. All rights are reserved, whether the whole or part of the material isconcerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publicationor parts thereof is permitted only under the provisions of the German Copyright Law of September 9,1965, in its current version, and permission for use must always be obtained from Springer. Violationsare liable for prosecution under the German Copyright Law.

Springer is a part of Springer Science+Business Mediaspringer.comc© Springer-Verlag Berlin Heidelberg 2008

The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,even in the absence of a specific statement, that such names are exempt from the relevant protective lawsand regulations and therefore free for general use.

Typesetting: by the authors and Integra, India

Printed on acid-free paper SPIN: 12038497

2007932286

Dr. Alias Abdul-Rahman Dr. Morakot PiloukESRI380 New York StreetRedlands [email protected]

Cover design: deblik, Berlin

5 4 3 2 1

Department of GeoinformaticsFaculty of GeoinformationScience and EngineeringSkudai [email protected]

Page 4: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Preface

This book is based on research works done by the authors at the University of Glasgow, Scotland, United Kingdom and the International Institute for GeoInformation Science and Earth Observation (ITC), The Netherlands in 2000 and 1996 respectively. We were motivated to write the book when we began a joint research work in 1992 for our postgraduate theses on Dig-ital Terrain Modelling (DTM) data structuring and eventually DTM soft-ware development based on triangular irregular network (TIN) data struc-ture. We realized then that many aspects needed to be addressed especially if an advanced geo information system (GIS) such as 3D GIS system was to be realized. Research in 3D GIS is getting growing in interest and this has really motivated us to do more experiments in the 3D domain. One of the most current interesting issues is spatial data modelling for 3D GIS.

We would like to thank our former supervisors, Dr Jane Drummond of University of Glasgow and Dr Klaus Tempfli of ITC. Various helps re-ceived from friends and colleagues at both institutions are also acknowl-edged. Special thanks go to Mohamad Hasif Nasaruddin, a postgraduate student at the Dept of Geoinformatics, Faculty of Geoinformation Science and Engineering, Universiti Teknologi Malaysia (UTM), Johor, Malaysia for his patient in formatting the manuscript.

This book aims to introduce a framework for spatial data modelling for 3D GIS and it is specifically written for GIS postgraduate level courses. Postgraduate students, researchers, and professionals in Geo Information (GI) science community may find this book useful and it may provide some insights in various spatial data modeling problems. We hope that this book will serve as one of the useful resources in 3D GIS or 3D geoinfor-mation research.

Alias Abdul-Rahman (UTM, Johor, Malaysia) Morakot Pilouk (ESRI, Redlands, CA, USA) 2007

Page 5: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Contents

Chapter 1 Introduction 1

1.1 Why Does 3D GIS Matter? 1 1.2 The Needs for 3D GIS 3 1.3 The Need for 3D Spatial Data Modeling 7 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13

Chapter 2 An Overview of 3D GIS Development 15

2.1 GIS Functions 15 2.2 3D GIS 16 2.3 Recent Progress Made on 3D GIS 17 2.4 Commercially Available Systems and 3D GIS 18 2.4.1 ArcView 3D Analyst 18 2.4.2 Imagine VirtualGIS 19 2.4.3 GeoMedia Terrain 20 2.4.4 PAMAP GIS Topographer 21 2.5 Why is 3D GIS Difficult to Realise? 22 2.6 Discussion 23

Chapter 3 2D and 3D Spatial Data Representations 25

3.1 Introduction 25 3.2 Classes of Object Representations 26 3.2.1 Grid 26 3.2.2 Shape Model 27 3.2.3 Facet Model 28 3.2.4 Boundary Representation (B-rep) 30 3.2.5 3D Array 32 3.2.6 Octree 33 3.2.7 Constructive Solid Geometry (CSG) 34 3.2.8 3D TIN (Tetrahedral network, TEN) 35 3.3 GIS Applicability of the Representations 37 3.4 The Selection Criteria 38 3.4.1 Representation of Object Primitives 38

Page 6: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

3.4.2 Topology of Spatial Objects: Simplexes and Complexes 40

3.5 Vector and Raster Representations 41 3.6 Summary 42

Chapter 4 The Fundamentals of Geo-Spatial Modelling 43 4.1 Spatial Data 44 4.2 Spatial Data Modeling 44 4.3 Models and Their Importance for Geoinformation 45 4.4 Components of Geo-spatial Model 47 4.5 Phases in Geo-spatial Modeling 48 4.6 Conceptual Design of a Geo-spatial Model 50 4.6.1 Definition of Space 51 4.6.2 Abstraction of Space 52 4.6.3 Abstraction of Real World Object 53 4.6.4 Object and Spatial Extent 57 4.6.5 Spatial Relations 57 4.6.6 Application of Spatial Relations 62 4.6.7 Representation of Spatial Objects and Relationships 65 4.6.8 Spatial Data Models in GIS 73 4.7 Logical Design of Geo-spatial Model 78 4.7.1 Relational Approach 79 4.7.2 Object-oriented Approach 81 4.8 Summary 85

Chapter 5 The Conceptual Design 87

5.1 TIN-based (2.5D) Data Model 87 5.2 Properties of the TIN-based Data Model 90 5.3 TEN-based Data Model 94 5.4 Generalized n-dimensional Integrated Data Model 97 5.4.1 The Definitions 98 5.5 Single-theme and Multi-theme 101 5.6 Euler’s Characteristics 102 5.6.1 Euler’s Equality 103 5.6.2 The Generalized Euler Equality 104 5.7 Discussion 107

VIII CONTENTS

Page 7: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Chapter 6 The Logical Design 109

6.1 Relational Approach 109 6.1.1 Relational Data Structure for TIN-based Model 110 6.1.2 Relational Data Structure for a TEN-based Model 112 6.1.3 Relational Data Structure for an n-dimensional Data Model 115 6.2 Object-oriented Approach 116 6.2.1 Object-oriented Definition of a Spatial Object 117 6.2.2 Object-oriented Design Based on IDM 118 6.2.3 Specialization of Classes 120 6.2.4 Aggregation of Objects 125 6.2.5 Creation of Objects 126 6.2.6 Behaviour of Objects in the Database 128 6.2.7 Comparison with Other OO Approaches 129 6.3 Discussion 130

Chapter 7 Object-Orientation of TIN Spatial Data 133

7.1 Introduction 133 7.2 Object-oriented Concepts 133 7.2.1 The Abstraction Mechanisms 134 7.2.2 The Programming Language 136 7.3 Object-oriented TIN Tessellations 136 7.3.1 Classes for 2D TIN Tessellations 136 7.3.2 Classes for 3D TIN Tessellations 140 7.4 Object-oriented Spatial Data Modelling 140 7.4.1 The Classes Schema 140 7.5 Object-oriented TIN Spatial Database Development 146 7.5.1 The POET OO DBMS 146 7.5.2 The POET Database Schema 147 7.5.3 The POET Database Browser 148 7.5.4 POET Database Query 148 7.6 Object-oriented TIN-based Subsystems for GIS 149 7.7 Summary 150

CONTENTS IX

S

TIN S

Page 8: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Chapter 8 The Supporting Algorithms 153

8.1 Introduction 153 8.2 Distance Transformation 153 8.3 Voronoi Tessellations 158 8.4 Triangulations (TINs) 163 8.4.1 TIN Topological Data Structuring 168 8.5 Visualization 170 8.6 3D Distance Transformation 171 8.7 3D Voronoi Tessellation 176 8.8 Tetrahedron Network (TEN) Generation 181 8.9 Constrained Triangulations 183 8.9.1 The Line Rasterization 183 8.9.2 The Construction of the Constrained TINs 185 8.10 Contouring Algorithm 190 8.10.1 Data Structures for Contouring 190 8.10.2 The Algorithm 192 8.10.3 The Contour Visualization 195 8.11 Algorithms for Irregular Network Formation 196 8.12 Summary 204

Chapter 9 Applications of the Model 207

9.1 Integration of Terrain Relief and Terrain Features 207 9.2 Creating an Integrated Database 209 9.3 A Spatial Query Example 212 9.4 Integrating with 3D Features 214 9.5 Integrating with Geo-scientific Data 219 9.6 Spatial Operators 221 9.7 Graphic Visualization 223 9.7.1 Wireframe Graphics 224 9.7.2 Hidden Line and Surface Removal 225 9.7.3 Surface Shading and Illumination 226 9.7.4 Texture Mapping 227 9.8 Virtual Reality 230 9.9 Discussion 230

X CONTENTS

Page 9: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Chapter 10 The Web and 3D GIS 233

10.1 Introduction 233 10.2 Web 3D GIS 234 10.3 Management of 3D Spatial Data 238 10.4 GUI for 3D Visualization and Editing on the Web 240 10.5 Current and Possible Approaches in Urban Planning 248 10.6 Realized Browser-based Solutions 249 10.7 Stand-alone Solutions/Toolkits/Front-ends 254 10.8 Summary 255

Chapter 11 Conclusion and Further Outlook 257

11.1 Summary 257 11.2 Further Research 264

References and Bibliography 267

Index 287

CONTENTS X I

Page 10: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Chapter 1 INTRODUCTION

1.1 Why does 3D GIS Matter?

Next generation of Geo Information System (GIS) requires a new way of spatial data modelling. We call the next generation of GIS 3D GIS. Fun-damentally, a new digital model has to be developed or established. Ex-ploiting digital computing technology to improve the quality of life, or to prevent or mitigate hazards or disasters, would first require the construc-tion of a model in digital form of the part of the earth and its environment. Such a model, a simplified description of complex reality, can conven-iently be used, stored, managed, maintained, distributed, and transported. Even a complex model may be stored on a small scale, in diskettes, tape cartridge or CD ROM, or transmitted via communication networks. A digital model contains spatial and non spatial aspects of reality and pro-vides a basis for operation and communication among the interested par-ties. A model distinguishes objects an object, or a set of objects, com-prises the elements of reality under investigation. Spatial aspects are those related to shape, size and location that pertain to geometric properties. Non spatial aspects include name, colour, function, price, ownership, and so forth, often referred to as thematic properties. Spatial aspects of reality can be well and economically represented in the form of graphics, whereas non spatial aspects, in many cases, can better be represented in text. Graphic representation facilitates rapid understanding of the situation in reality, permitting high level abstraction or description about neighbouring rela-tionships, while the textual representation is more suitable for aspects that cannot be graphically described. A digital model must be capable of relat-ing these two representations. Creating such a model as an artificial con-struction of reality in a computing environment requires a tool set exploit-ing the technology both of computer graphics (CG) (Sutherland, 1963, 1970; Foley et al., 1992; Watt, 1993) and database management (DBMS). Geographic information systems (Burrough, 1986; Maguire et al., 1991), and computer aided design (CAD) are examples of such tools. The essen-tial difference between GIS and CAD is the handling of the spatial aspects rather than the non spatial aspects.

Geographical Information Systems (GISs) represent a powerful tool for capturing, storing, manipulating, and analysing geographic data. This tool is being used by various geo-related professionals, such as surveyors, car-tographers, photogrammetrists, civil engineers, physical planners (urban and rural), rural and urban developers, geologists, etc. They use the tool

Page 11: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

2 Chapter 1

for analysing, interpreting, and representing the real world and understand-ing the behaviour of the spatial phenomena under their respective jurisdic-tions. Almost all of the systems used by the geoinformation community to date are based on two-dimensional (2D) or two-and a half-dimensional (2.5D) spatial data. In other words, one may find difficulty processing and manipulating spatial data of greater dimension than 2 in the existing sys-tems, resulting in inaccurate or at least very incomplete information. Fur-thermore, manipulating and representing real world objects in 2D GIS with relational databases are no longer adequate because new applications de-mand and increasingly deal with more complex hierarchical spatial data than previously supported by the relational model. It has been suggested in the literature that the abstraction of complex spatial data could be han-dled more effectively in object-oriented rather than in relational database environment (Egenhofer and Frank, 1989; Worboys, 1995).

The limitations of the current 2D GISs, especially in geoscience, have been reported in the literature by Jones (1989), Raper and Kelk (1991), RongxingLi (1994), Houlding (1994), Bonham-Carter (1996), and Wei Guo (1996). The limitations mentioned relate to data dimensionality and data structures. Single valued z-coordinate data such as a point (x, y coordi-nates) with the z-coordinate representing height presents no data handling difficulty in such systems, but it is inadequate for data with multiple z-values (Bonham-Carter, 1996; Raper and Kelk, 1991) such as ore bodies and other important three-dimensional real world entities. A major im-pediment to establishing 3D GISs is associated with inappropriate spatial data structures, as reported in Jones (1989) and Rongxing Li (1994). These two authors have proposed voxel data structures for 3D data as a so-lution to the data structuring problem, but no real operational system was developed based on the structure. The problem was also highlighted in the geological field by Houlding (1994). True representations and spatial in-formation, for example sub-surface 3D objects, could not be successfully achieved with 2D systems. 3D visualisation tools alone (for example Ad-vanced Visualization System (AVS), Voxel Analyst of Intergraph, and other Digital Terrain Model (DTM) packages) were not able to truly man-age such data as demanded. For example Wei Guo (1996) experimented with the 3D modelling of buildings by using Molenaar’s (1992) formal data structure in the relational database environment together with Auto-Cad as a 3D visualization tool; AutoCad was used to generate the building models. In the literature, a common suggestion has been that the existing GISs were able to handle most of the 2D spatial data, but had difficulty in handling 3D spatial data and beyond, therefore, an extension of the existing

Page 12: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 3

systems to at least a third-dimension (3D) is one of the solutions suggested by GIS researchers.

Another observation is that the literature cites no work on three-dimensional GIS coupled with object-oriented technology. Given that the weakness of conventional off-the-shelf 2D or 2.5D GISs are revealed when three-dimensional real world entities are considered, it is understood that object-orientation and three-dimensionality are not more often jointly con-sidered. Some works have focussed on 3D issues such as work reported in Fritsch and Schmidt, 1995; Kraus, 1995; and Fritsch, 1996. But all of these attempts were based on the relational database environment. There-fore, this research monograph looks at both 2D and 3D spatial data model-ling and the development of a geoinformation system using relational and object-oriented technology to attempt to solve 3D problems in the GIS en-vironment.

1.2 The Need for 3D GIS

We live in a three dimensional (3D) world. Earth scientists and engineers have long sought graphic expressions of their understanding about 3D spa-tial aspects of reality in the form of sketches and drawings. Graphical de-scriptions of 3D reality are not new. Drawings in perspective view date from the Renaissance period (Devlin, 1994). 3D descriptions of reality in perspective view change with the viewing position, so their creation is quite tedious. Traditional maps overcome this problem by using orthogo-nal projections of the earth. However, they offer a very limited 3D impres-sion.

These traditional drawings and maps reduce the spatial description of 3D objects to 2D. Using computing technology, however, knowledge about reality can be directly transferred into a 3D digital model by a process known as 3D modelling. A 3D description of reality is independent of the viewing position. Adequate cover of the aspects of reality under investiga-tion requires its understanding from many different viewpoints. The disci-plines of geology (Carlson, 1987; Bak and Mill, 1989; Jones, 1989; Youngman, 1989; Raper and Kelk, 1991), hydrology (Turner, 1989), civil engineering (Petrie and Kennie, 1990), environmental engineering (Smith and Paradis, 1989), landscape architecture (Batten, 1989), archeology, me-teorology (Slingerland and Keen, 1990), mineral exploration (Sides 1992), 3D urban mapping (Shibasaki et al., 1990; Shibasaki and Shaobo, 1992), all draw on 3D modelling for the efficient completion of their tasks.

Page 13: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

4 Chapter 1

A 3D model is the basis of a system providing the functionality to accom-plish the task in hand. Scott (1994) has summarized the work of Bak and Mill (1989), Fisher (1993), Kavouras and Masry (1987), Raper (1989), Raper and Kelk (1991), and Turner (1989), to provide a set of functions that can be expected from 3D modelling. These studies should provide the means for constructing a 3D model from disparate inputs; permit the main-tenance of existing models; facilitate effective 3D visualization with, for example, orthographic, perspective or stereo display with hidden line/surface removal, surface illumination, texture mapping; spatial analy-ses enabling the calculation of volume, surface area, centre of mass, opti-mal path as well as spatial and non spatial search and inquiry.

CAD is a typical CG tool for 3D modelling used in car, machinery, aircraft and spacecraft designs, the construction industry, and architecture. CAD focuses on the geometric aspect of the model and its 3D visualization. An example would be a perspective view with hidden line and surface re-moval, surface illumination, ray tracing, and texture mapping. The ques-tion arises whether CAD can support all the tasks required in the disci-plines listed above. Attempts have been made to use CAD for tasks in earth sciences requiring 3D modelling and functionality. However, it can-not immediately be assumed that CAD is suited to these tasks, for the fol-lowing reasons.

CAD was developed to solve problems in the design of man made ob-jects with well or predefined shapes, sizes, spatial relationships and thematic properties. CAD does not provide the tools for data structur-ing, or dealing with objects lacking such well-defined shapes, sizes, spatial relationships and thematic properties. Neither is it capable of analysing spatial relationships, nor coping with the disparate data sets and uncertainty typically encountered in GIS. For example, CAD will not reliably maintain the neighbourhood relationships between objects important in earth science analyses, because these relationships may not be considered significant in the design.

Designing an object, such as a building, is a subjective matter. All as-pects of objects and their relationships have to be decided by a human designer; there is little that can be automated. Earth science applica-tions seek to model existing objects, with shapes, sizes and interrela-tionships outside human control. Here, automation is desirable because of the large number of objects involved. Some relationships important for spatial analysis have to be created automatically. CAD does not usually provide a function for this kind of automation.

Page 14: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 5

CAD starts the object definition from 3D. When objects are broken down in 2D components, the relationships between them are known. Earth science applications typically model components of reality sepa-rately, mostly in 2D, and are dominated by the application view, avail-able tools and information. The components have to be combined and their interrelationships discovered at a later stage. This is quite diffi-cult, since CAD does not usually provide sufficient tools to derive the relationships between the separate components.

CAD creates a complex object by combining several components pos-sessing such simple geometry as a cube, cylinder, or sphere. The op-erations of transformation, union, and intersection can be readily ap-plied to such components to obtain the complex object. Earth science applications usually treat a complex object as a whole. Decomposition into primitives is comparable to reverse engineering, the opposite of CAD. The modelling approach used by CAD may not therefore always be suitable for earth science applications. Geometric primitives of an even lower level, such as points and lines, are needed to represent complex reality beyond man made objects.

These geometric primitives also determine the related operations which CAD may not be capable of providing.

A more suitable tool for earth science applications would be a GIS provid-ing a 3D modelling capability, that is to say, a 3D GIS. At the time of writ-ing, a GIS capable of providing the functions listed above list with full 3D modelling capability is not commer-cially available. Most GISs still limit their geometric modelling capability to 2D so that the 3D representation, analysis and visualization provided by CAD are not possible. Most en-deavours to model the third dimen-sion can be found in the representa-tion of terrain relief and in digital terrain models (DTM). DTM can fa-cilitate spatial analyses related to re-lief, including slope, aspect, height zone, visibility, cut and fill volume, and surface area, and the 3D visuali-zation of a surface, as in a perspective view. However, the basis of DTM is a continuous surface with a single height value for every planimetric

(a)

(b)

(c)

Fig. 1.1 Single-valued surface (a), 3D solid object (b) and multi-valued sur-face (c).

Page 15: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

6 Chapter 1

location (see Figure 1.1a). DTM cannot accommodate a 3D (solid) object, or a surface with multiple height values at a given planimetric location (see Figure 1.1b and Figure 1.1c, respectively).

Although raster-based systems which could be regarded as 3D GISs are available, they may not be able to maintain the knowledge about reality available in the original data set. This knowledge may be lost because of problems in resolution and resampling. As a remedy, the original data set would have to be stored separately from the model, for example, for: • recreating the model if the result proves to be unsatisfactory because of

unsuitable mathematical definition • creating another model with different resolution • merging with another data set to create a new model • archiving as a reference to, or evidence of, the model.

These activities imply the need to store original data in an appropriate structure ready for future use. Necessary information about the data should be attached to each data element. In DTM for instance, information that a line is a breakline should be kept because it will have an impact on the in-terpolation. Similarly, other information can be attached which influences data handling strategies.

Since neither CAD nor GISs can at present fulfil the requirements of earth science applications, further research and development of a 3D GIS would seem appropriate.

Who needs 3D GIS?

As in the popular 2D GIS for 2D spatial data, 3D GIS is for managing 3D spatial data. Raper and Kelk (1991), Rongxing Li (1994), Förstner (1995), and Bonham-Carter (1996) present some of the three dimensional applica-tion areas in GIS, including:

• ecological studies • environmental monitoring • geological analysis • civil engineering • mining exploration • oceanography • architecture • automatic vehicle navigation • archaeology

Page 16: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 7

Objects with known or well-defined spatial ex-tent, location and prop-erties

Objects with unknown or not well-defined spa-tial extent, location and properties

Fig. 1.2 Two types of real world objects with respect to their spatial extent.

• 3D urban mapping • landscape planning • defence and intelligence • command and control

The above applications may pro-duce much more useful information if they were handled in a 3D spatial system, but 3D spatial objects on the surface and subsurface appear to demand more complex solutions (e.g. in terms of modelling, analysis, and visualization) than the existing systems can offer.

1.3 The Need for 3D Spatial Data Modelling

In addition to the problem of creating a system capable of offering 3D modelling and functionality, there is a further problem concerning the type of 3D model chosen as the basis for 3D GIS. The model contains knowl-edge about reality, so we consider below the types of real world objects it must represent. Two kinds of real world objects may be differentiated in terms of prior knowledge about their shapes and location, as shown in Fig-ure 1.2. In reality, objects from the two categories coexist. Traditional GIS models the objects of each category independently with the result that two separate kinds of systems or subsystems have been developed.

Raper (1989) has also defined these two categories of objects. The first category, regarded as ‘sampling limited’, is for objects having discrete properties and readily determined boundaries, such as buildings, roads, bridges, land parcels, fault blocks, perched aquifers. The second category, known as ‘definition limited’, is for objects having various properties that can be defined by means of classification, using property ranges. For ex-ample, soil strata may be classified by grain-size distribution; moisture content, colloid or pollutant in the water by percentage ranges; carbon monoxide in the air by concentration ranges, and so forth. Molenaar (1994a) regards these objects as ‘fuzzy spatial objects’.

Separate modelling of these two categories of objects tends to contradict the reality, which leads to difficulties in representing their relationships. Such a question as, ‘how many of the people working in a 50-storey office

Page 17: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

8 Chapter 1

building are affected by polluted air generated by vehicles in nearby streets during rush hours?’ cannot be answered until the two separate models are combined, as shown in Figure 1.3. Modelling them together with more accurate represen-tation of their relationships in the 3D environment requires the integrated 3D modelling.

Note also that the properties of an object may be well de-fined in some specific dimen-sions and ill defined in others. For example, given a DTM data set representing a sur-face, the planimetric extent of regions at the elevation of 100 metres above mean sea level cannot be defined until the re-sult of interpolation based on a mathematical definition (for example, lin-ear interpolation) is obtained. That is to say, although the spatial extent of this region may be known in the z-dimension, the spatial extent in plani-metry (x, y) has still to be discovered. The model must contain the aspect allowing the appropriate operation, such as interpolation or classification, if the required description of the properties of an object is to be obtained.

Apart from the problem of the separate modelling of the two types of ob-jects, there remains the further problem of the separate modelling of an ob-ject’s components. These components are relief and planar geometry asso-ciated with thematic properties. This separation has resulted in independent systems and data structures, DTM and 2D GIS, respectively. The consequences are data redundancy, which may lead to uncertainty when the two data sets are combined and only one data set has been up-dated.

DTM can facilitate several GIS analyses and visualization taking into ac-counts the third dimension. The spatial information stored in DTM and in GIS, however, can only be related through coordinates. This implies that relationships between different components may not be properly repre-sented because of metric computation instead of topology. To overcome this, information derived from DTM must be converted into a form GIS can recognize. For example, information about a slope or height zone must first be converted into a thematic layer of GIS for further overlaying before

Fig. 1.3 An example of two types of real world objects

Page 18: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 9

the spatial analysis can be carried out. Imagine having information about the relief, planimetry and themes integrated into one model, so that con-version of such information as slope, height zone and so forth were no longer necessary. Such a question as, ‘which land parcels are subject to one-metre flooding?’ could be answered from one model. Integrated mod-elling of this kind is evidently also required for 3D GIS.

1.4 Problems Associated with Spatial Modelling

Establishing a 3D GIS while taking into account the integration of the nec-essary components and different types of objects requires the solution of the following problems related to the spatial model representing reality:

1) Design of a spatial model • design of an integrated data model, or a scheme, permitting the deri-

vation of a unified data structure capable of maintaining all the com-ponents of the geometric representation of real world objects, whether obtained from direct measurements or from derivations, in the same database. Each geometric component must be capable of representing a real world object differently understood by different people.

2) Construction of a spatial model • development of appropriate means and methods for 3D data acquisi-

tion; • coordinate transformation into common georeferencing when differ-

ent components are to be included into one database; • development of a data structuring method that unites the data from

various inputs of multi sources into an integrated database capable of being maintained by a single database management system;

• design of thematic classes to organize representation of real world ob-jects with common aspects into the same category;

• solving the uncertainty arising from discrepancies from different data sets during the integration process and converting the uncertainty into a ‘data quality’ statement to be conveyed to the end user.

3) Utilization of a spatial model • utilization of existing components, such as 2D data and DTM (back-

ward compatibility) and preparation of those components for future incorporation into the higher-dimension model (forward compatibil-ity) to save the costs of repeating data acquisition.

for 3D GIS

Page 19: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

10 Chapter 1

• development of additional spatial operators and spatial analysis func-tions;

• development of maneuverable graphic visualization permitting the se-lection of appropriate viewpoints and representation enabling conven-ient, adequate uncovering of the details of objects stored in the data-base;

• design of 3D cartographic presentation of information, including name placement, symbol, generalization, etc.;

• design of a user interface and query language allowing users access to the integrated database;

• development of a spatial indexing structure that speeds up data re-trieval and storage processes for the integrated database, including specific (database) views for each user group and guidelines keeping these views updated according to the core database;

• development of tools for navigating among different models stored in databases at different sites and computing platforms.

4) Maintenance of spatial model • design of updating procedures, including the development of consis-

tency rules ensuring the logical consistency and integrity of the inte-grated database, especially during the updating process.

1.5 Previous Work

The status and progress of research in the 3D GIS field within the scope of this monograph and the identification of solutions and remaining problems are made clear from the following review of previous work.

The development of data models for a 3D GIS has branched in two direc-tions. The first is the full 3D approach that looks directly into the design of a data model suitable for 3D GIS. Molenaar (1989) proposes a formal data structure (FDS) for a 3D vector map which may be regarded as a generali-zation of the 2D version of FDS. Shibasaki and Shaobo (1992), Rikkers

(1994), and Wang (1994) have reported experimental use of 3D FDS.

The second approach comes from the viewpoint referred to as the ‘integra-tion of DTM and GIS’. DTM became a discipline in its own right in the late 1950s (Miller and Laflamme, 1958). Fritsch (1990) has recognized the work of Makarovic (1977) as a proposer of this integration. Males (1978) though not addressing the integration issue, demonstrated the use of a

et al. (1993), Bric (1993), Bric et al.,

Page 20: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 11

triangulated irregular network (TIN) permitting the attachment of thematic information with elements of TIN in the ADAPT system.

Further steps towards this integration date from the late 1980s, when DTM became an essential part of many complex spatial analyses in GIS in ero-sion and slope protection, flood protection, the planning of irrigation for agriculture, the geometric correction of remotely sensed images, and so forth. Würländer (1988) investigates some strategies for integrating DTM into GIS. Sandgaard (1988) describes an attempt at integrating DTM into the Dangraf system to facilitate the production of maps with contour lines. Mark and colleagues (1989) report an approach to interfacing a GIS based on quadtree (Samet 1990) with a regular grid DTM for display or analysis. Ebner and colleagues (1990) propose the ‘subroutine interface’ which was implemented in the program package HIFI-88. Subroutines for interactive editing of GIS are provided for updating DTM, for example, point inser-tion and deletion, and the change of coordinates in planimetry and height while databases of DTM and GIS remain separate. Ebner and Eder (1992) report drawing on this approach to the facilitation of spatial analysis, using the HIFI-GIS interface with the SICAD-Hygris System to analyse forest damage in terms of such relief parameters as height, slope and exposition. Fritsch (1990) reports the realization of integration at the data structure level. Rather than a full 3D data structure, he suggests an approach that separates two geometric databases for terrain and situation data from an-other for thematic data. These three data sets are managed within one ob-ject oriented database environment. Fritsch and Pfannenstein (1992a) weigh the advantages and disadvantages of integration based on regular-grid, TIN and a hybrid of both. Fritsch and Pfannenstein (1992b) extend this comparison to the layer (organizing different themes in specific layers) and object class (organizes objects into a hierarchy) approach.

An issue in spatial modelling concerns the representation of spatial rela-tionships. Egenhofer (1989), Jackson (1989), Kainz (1989), and Pigot (1991) have described the representation of spatial relationships between objects in 2D and 3D space, based on sound mathematical concepts.

Regarding the issue of model construction, CAD and most CG software packages provide interactive tools for the manual construction of models of objects with discernible boundaries. Manual construction is labourious and the method would not cope with large numbers of objects. For objects with indiscernible boundaries, significant progress has been made in com-putational geometry based on 2D and 3D Voronoi tessellation (Voronoi 1908, Thiessen 1911, Dirichlet 1850), in the construction of TINs, and tet-rahedral networks (TEN). Watson (1981), Avis and Bhattacharya (1983),

Page 21: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

12 Chapter 1

Edelbrunnner and colleagues (1986), Tsai and Vonderohe (1991), Midtbø (1993) have all suggested methods for the construction of TEN based on Delaunay triangulation criteria (Delaunay 1934). These methods were ex-tensively applied long ago to the construction of TIN (Shamos and Hoey 1975, Lawson 1977, Lewis and Robinson 1978, Sibson 1978, McCullagh and Ross 1980, Lee and Schachter 1980, Bowyer 1981, Watson 1981, Mirante and Weingarten 1982, Maus 1984, Dwyer 1987, Sloan 1987,

ver, these developments are quite independent of GIS.

For the issue of the exploitation of the 3D model, considerable progress has been reported in two other disciplines exploiting CG technology, namely CAD and virtual reality (VR). CAD and VR provide a realistic visualization capability, that is to say, perspective display with hidden line and surface removal, shading and surface illumination, ray tracing, and texture mapping. In addition, VR provides high interactivity within the concept of ‘functional realism’, allowing the user to manipulate and inter-act with virtual objects stored in the computer’s database as in reality. For instance, the user can ‘grab’ a virtual object displayed on the computer screen, using the interfacing device called a ‘data glove’ which sends feed-back to the user’s hand (for example, a pulse, or vibration) as soon as the virtual object is virtually touched. Developments in this direction are also quite independent of GIS.

The status of the research in 3D GIS and the most relevant remaining prob-lems can be summarized in the following statements:

• The full 3D approach, 3D FDS, does not support well the modelling of real world objects whose boundaries cannot be directly deter-mined. Further extension to cover this issue is therefore needed.

• Progress made by the integration approach can only achieve solutions for surface related objects with little support from theoretical concept of spatial modelling. Extension of this approach to full 3D based on sound spatial mathematics is required.

• Efficient methods for data acquisition, data structuring, database crea-tion and updating with respect to 3D GIS have yet to be developed.

• The incorporation into 3D GIS of independent developments in 3D visualization and 3D geometric construction, whether manual (inter-active 3D graphical editing) or automatic (3D Voronoi and tetrahedral network), needs further research.

Macedonio and Pareschi 1991, etc.). Howe

Page 22: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

INTRODUCTION 13

1.6 Background to the 3D GIS Problem

In geomatics or geoinformatics we consider real world objects exist in three-dimensional (3D), thus it is desirable to have a system which is able to store, handle, manipulate, and analyse objects in a 3D environment. As mentioned in the previous section, the current popular GIS software han-dles, manipulates, and analyses geographic data in 2D or 2.5D, thus using this system to manipulate 3D data full (particularly multiple Z coordinates) information about real world objects may not be appropriate. Therefore, the 2D GIS (or 2.5D GIS) needs to be extended, i.e. to 3D GIS. Only within the last decade has 3D GIS begun to be discussed in the GIS re-search community (Raper and Kelk, 1991; Rongxing Li, 1994). The de-velopment of this particular GIS approach seems to be relatively slow due to the lack of proper spatial data models and data structures, and the lack of a comprehensive theory of object relationships and data basing for the 3D environment (Wei Guo, 1996). Attempts have been made to develop 3D GIS by Li et al. (1996), Pilouk (1996) and Qingquan Li and Deren Li (1996). Li’s use an octree approach for 3D subsurface geological model-ling, Pilouk uses a 3D TIN approach for regular features on the terrain, while a combination of octree/tetrahedron was proposed by Qingquan Li and Deren Li. Others have used Constructive Solid Geometry (CSG) and Boundary-representation (B-rep) approaches (Cambray, 1993; Cambray and Yeh, 1994; Bric, 1993; Bric et al, 1994; and Zeitouni et al, 1995). All of this work were based on regular shaped objects, which were man-made, and relational data basing. Nonetheless, there appears very little published work on the modelling of 3D objects including natural objects, e.g. forests, plants, water bodies, and other natural subsurface features using the object-oriented (OO) approach. Recent research (Rongxing Li, 1994 and more recently Fritsch, 1996) in this domain have suggested that 3D spatial data modelling, structuring and data basing with object-orientation leads to bet-ter 3D GIS. This suggestion seems mainly arise from the complexity of 3D spatial data, as well as some positive features of object-orientation where every physical or spatial object of the real world can be defined dur-ing software development. It is therefore imperative to investigate the practicality of a means to improve the representation of natural objects in 3D and to manage them in an object-oriented GIS.

Page 23: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

Chapter 2 AN OVERVIEW OF 3D

The previous chapter has introduced the importance and some of the exist-ing problems in 3D spatial data modelling and in developing an informa-tion system based on 3D spatial data. In this chapter, several types of two-dimensional (2D) GIS systems which are related to the development of 3D GIS will be further discussed. Some well established systems which are currently available in the market will be reviewed. Since data structures, data modelling and database management are important aspects of system development, all the discussions and system overview will focus on these aspects.

2.1 GIS Functions

Any GIS system should be able to provide information about geo spatial phenomena. Principally, the tasks or the functions of a GIS system are to: 1) capture, 2) structuring, 3) manipulation, 4) analysis, and 5) presentation (Raper and Maguire, 1992).

• Capture. Capturing is inputting spatial data to the system. Many dif-ferent techniques and devices are available for both geometric and at-tribute data. The devices in frequent use for collecting spatial data can be classified as manual, semiautomatic or automatic, and the output either in vector or raster format. Detailed discussion on data captur-ing is not covered here.

• Structure. Structuring is a crucial stage in creating a spatial database using GIS. This is because it determines the range of functions which can be used for manipulation and analysis. Different system may have different structuring capabilities (simple or complex topology, relational or object-oriented).

• Manipulate. Among important manipulation operations are generali-sation and transformation. Generalisation is applied for smoothing spatial data and it includes line smoothing, points filtering, etc. Transformation includes among others coordinate transformation to a specified map projection and scaling.

• Analysis is the core of a GIS system. It involves metric and topologi-cal operations on geometric and attribute data. Primarily, analysis in

GIS DEVELOPMENT

Page 24: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

16 Chapter 2

GIS concerns operations on more than one set of data which gener-ates new spatial information of the data. Terrain analysis (e.g. inter-visibility), geometric computations (volume, area, etc), overlay, buff-ering, zoning are among typical analysis functions in GIS.

• Presentation is a final task in GIS. At this stage, all generated infor-mation or results will be presented in the form of maps, graphs, ta-bles, reports, etc.

Ideally, a 3D GIS should have the same functions as a 2D GIS. However, such 3D systems are not available due to several impediments. The ensu-ing sections will discuss the challenges in 3D GIS development.

2.2 3D GIS

In this section, some problems and related issues in 3D GIS software de-velopment are reviewed and discussed. 3D GIS should be able to model, represent, manage, manipulate, analyse and support decisions based upon information associated with three-dimensional phenomena (Worboys, 1995). The definition of 3D GIS is very much the same as for 2D system. In GIS, 2D systems are common, widely used and able to handle most of the GIS tasks efficiently. The same kind of system, however, may not be able to handle 3D data if more advanced 3D applications are demanded (Raper and Kelk, 1991; Rongxing Li, 1994) such as representing the full length, width and nature of a borehole (some examples of 3D applications areas are listed in section 2.3). 3D GIS very much needs to generate in-formation from such 3D data. Such a system is not just a simple extension by another dimension (i.e. the third dimension) on to 2D GIS. Adding this third dimension into existing 2D GIS needs a thorough investigation of many aspects of GIS including a different concept of modelling, represen-tations and aspects of data structuring. Existing GIS packages are widely used and understood for handling, storing, manipulating and analysing 2D spatial data. Their capability and performance for 2D and for 2.5D data (that is also DTM) are generally accepted by the GIS community. A GIS package which can handle and manipulate 2D data and DTM cannot be considered as a 3D GIS system because DTM data is not real 3D spatial data. The third dimension of the DTM data only provides (often after in-terpolation) a surface attribute to features whose coordinates consist only of planimetric data or x, y coordinates. GIS software handling real 3D spatial data is rarely found. Although the problem has been addressed (as mentioned in chapter one) by several researchers such as Raper and Kelk

Page 25: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

AN OVERVIEW OF 3D GIS DEVELOPMENT 17

(1991), Cambray (1993), Rongxing Li (1994), Pilouk (1996), and Fritsch (1996), some further aspects particularly spatial data modelling using rela-tional and OO techniques need to be investigated. This modelling issue will be addressed in later chapter.

2.3 Recent Progress Made on 3D GIS

Some recent research efforts by the GIS community has focussed on how to develop 3D systems; data structures and data models are major aspects of GIS system development. These efforts are summarised below.

Much previous work done on 3D data modelling concentrated on the use of voxel data structures (Jones, 1989). This particular approach does not address spatial modelling aspects (that is also topological aspect of the data); it is only useful for the reconstruction of 3D solid objects and for some basic geometric computations. Another problems with this data model is that it needs very large computer space and memory.

Carlson (1987) has proposed a model called the simplicial complex. He uses the term 0-simplex, 1-simplex, 2-simplex, and 3-simplex to denomi-nate spatial objects of node, line, surface, and volume. His model can be extended to n-dimensions.

Cambray (1993) has proposed CAD models for 3D objects combined with DTM as a way to create 3D GIS, that is a combination of Constructive Solid Geometry (CSG) and Boundary representation (B-rep).

Other attempts to develop 3D GIS can be found in Kraus (1995), Fritsch and Schmidt (1995), and Pilouk (1996). These attempts were based on the TIN data structure to represent 3D terrain objects but no report exists on any related aspects of using OO techniques for modelling and data struc-ture.

Data modelling and structuring of 3D spatial objects in GIS has not been as successfully achieved as in CAD (Li, 1994). Data modelling in GIS is not only concerned with the geometric and attribute aspects of the data, but also the topological relationships of the data. The topology of spatial data must be available so that the neighbouring and connectedness between ob-jects can be determined. There are a number of mathematical possibilities for the determination of the topological description of objects. The infor-mation gained from the generated TIN neighbours is useful for further spa-tial analysis and applications. Topological relationships for linear objects as represented by TIN edges can be established. One edge is represented

Page 26: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

18 Chapter 2

by a start node and an end node. From this edge topology, a chain of edges or arcs could be easily established. For TIN data, another approach is the simplicial complex developed by Carlson. A TIN’s node is equiva-lent to 0-simplex, TIN’s edge is equivalent to 1-simplex, a TIN surface (area) is equal to 2-simplex, and 3-simplex is equivalent to a 3D TIN (tet-rahedron). The simplicial complex technique checks the consistency of generated TIN structures by Euler’s equality formulae (see Carlson (1987) for a detailed discussion). An OO TIN approach is described in later

2.4 Commercially Available Systems and 3D GIS

There are few systems available in the market which can be categorised as a system which attempts to provide a solution for 3D representation and analysis. Four systems are chosen for detailed consideration. They were chosen because they constitute a large share of the GIS market and provide some 3D data processing functions. The systems are the 3D Analyst of ArcView (from Environmental System Research Institute or ESRI Inc.), Imagine VirtualGIS (from ERDAS Inc.), GeoMedia Terrain from Inter-graph Inc. and PAMAP GIS Topographer. The following review is based on the available literature and Web-based product reviews.

2.4.1 ArcView 3D Analyst

The 3D Analyst (3DA) is one of the modules available in ArcView GIS. In ArcView these modules are known as extensions. The system’s exten-sions and the main GIS module, that is the ArcView itself, is shown in Figure 2.1. ArcView is designed to provide stand alone and corporate wide (using client-server network connectivity) integration of spatial data (Maguire, 1999). The 3DA can be used to manipulate 3D data such as 3D surface generation, volume computation, draping for other raster images (such Landsat TM, SPOT, GeoSPOTV images, aerial photos or scanned maps), and other 3D surface analysis functions such as terrain intervisibil-ity from one point to another (ESRI, 1997).

chapter.

Page 27: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

AN OVERVIEW OF 3D GIS DEVELOPMENT 19

Fig. 2.1 The 3D Analyst (shown on top of the extension’s box) within ArcView system

The system runs mainly on personal computers and accepts several operat-ing system such Windows 95/98/2000 and Windows NT 4.0 as well as wide range of UNIX platforms (ESRI, 2000). The system works mainly with vector data. Even though raster files can be incorporated into 3DA, it is only for improving the display of vector data (e.g. by draping vector data with aerial photo images). (Raster files are and considerably for aspect of 2-D spatial data analysis.)

In summary, 3DA can be used to manipulate 3D data especially for visu-alization purposes. Thus, ArcView is very much a 2D GIS system, but 3DA supplies 3D visualization and display (e.g. of data with x, y, z coor-dinates). 3D GIS analysis is not achieved. It is worth noting, however, that 3DA supports triangular irregular network (TIN) data structure.

2.4.2 Imagine VirtualGIS

The Imagine system was originally developed for remote sensing and im-age processing tasks. Recently, the system has provided a module for GIS. The Imagine system is one of the GIS solutions developed by ERDAS Inc

ArcView

Spatial database

GIS Functions

Extensions

3D Analyst

Spat ial Analyst

User Interface

Image Analysis

Tracking Analyst

Internet Map Server

Business Analyst

Network Analyst

Street Map

Street Map 2000

ArcPress

Windows OS UNIX OS

Core system

Page 28: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

20 Chapter 2

(ERDAS, 2000). The GIS module is called VirtualGIS. It is a module that provides a three-dimensional visual analysis tools. The system has run under various computer systems ranging from personal computers to workstations such as DEC computers, IBM personal computers, Hewlett Packard, Sun Sparc and IBM RISC machines. Currently, the system works with operating systems such as Windows98/2000, Windows NT and various UNIX systems. It is a system which has an emphasis on dynamic visualisation and real-time display in the 3D display environment. Besides various and extensive 3-D visualizations, the system also provides fly-through capabilities (Limp, 1999). Figure 2.2 shows the system overview of the VirtualGIS with its core Imagine system.

Fig. 2.2 The VirtualGIS component (shown on top of the Add-on module’s box) in the Imagine system architecture.

As with 3DA, this system also centres around 3D visualization with true 3D GIS functions hardly available.

2.4.3 GeoMedia Terrain

GeoMedia Terrain is one of the subsystems that work under the Geo-Media GIS system developed by Integraph Inc. The system runs under the Windows operating systems (including NT 4.0 system). The Terrain

SpatialDatabase

IMAGINE

Image

Processing

VirtualGIS

User Interface

Developer Tool Kit

SubPixel Classifier

Vector

NITF

ATCOR2

Add-on Module

Core system

DEC OS Windows IBM RISC SGI OS SUN OSOS

Page 29: Spatial Data Modelling for 3D GIS · 1.4 Problems Associated with Spatial Modelling for 3D GIS 9 1.5 Previous Work 10 1.6 Background to the 3D GIS Problem 13 Chapter 2 An Overview

AN OVERVIEW OF 3D GIS DEVELOPMENT 21

system performs three major terrain tasks, namely, terrain analysis, terrain model generations, and fly-through (Integraph, 2000). In general, the Ter-rain serves as DTM module for the GeoMedia GIS as with other systems mentioned in the previous sections where true 3D GIS capabilities are hardly offered by software vendors. Figure 2.3 shows the Terrain subsys-tem within the GeoMedia core system.

Fig. 2.3 The Terrain component within the GeoMedia system

2.4.4 PAMAP GIS Topographer

This GIS system is one of PCI Geomatics Inc.’s products. It runs under Windows95/98 and NT operating systems. PAMAP GIS is a raster and vector system (Geomatics, 2000). Besides its 2D GIS functions, the sys-tem has a module for handling 3D data, called Topographer as depicted in Figure 2.4. Four main GIS modules are offered, they are Mapper, Model-ler, Networker and Analyser which form the core system. For 2D data handling, the system performs GIS tasks as in other systems mentioned earlier. For 3D data, most of the 3D functions in the Topographer work as by any DTM packages, for example terrain surface generation, terrain sur-faces analysis (e.g. calculation of area, volume) and 3D visualisation (such as perspective viewing). This system also focuses on 3D display of terrain data.

SpatialDatabase

User Interface

Add-on Module

Core system

Windows NT OSOS

GeoMediaGIS

Functions

SpatialDatabase

Terrain